1--TEST-- 2mb_stripos() 3--EXTENSIONS-- 4mbstring 5--FILE-- 6<?php 7// TODO: Add more encodings 8 9ini_set('include_path','.'); 10include_once('common.inc'); 11 12// Test string 13$euc_jp = "0123\xA4\xB3\xA4\xCE\xCA\xB8\xBB\xFA\xCE\xF3\xA4\xCF\xC6\xFC\xCB\xDC\xB8\xEC\xA4\xC7\xA4\xB9\xA1\xA3EUC-JP\xA4\xF2\xBB\xC8\xA4\xC3\xA4\xC6\xA4\xA4\xA4\xDE\xA4\xB9\xA1\xA30123\xC6\xFC\xCB\xDC\xB8\xEC\xA4\xCF\xCC\xCC\xC5\xDD\xBD\xAD\xA4\xA4\xA1\xA3"; 14 15$slen = mb_strlen($euc_jp, 'EUC-JP'); 16echo "String len: $slen\n"; 17 18// EUC-JP - With encoding parameter 19mb_internal_encoding('UTF-8') or print("mb_internal_encoding() failed\n"); 20 21echo "== POSITIVE OFFSET ==\n"; 22 23print mb_stripos($euc_jp, "\xC6\xFC\xCB\xDC\xB8\xEC", 0, 'EUC-JP') . "\n"; 24print mb_stripos($euc_jp, '0', 0, 'EUC-JP') . "\n"; 25print mb_stripos($euc_jp, 3, 0, 'EUC-JP') . "\n"; 26print mb_stripos($euc_jp, 0, 0, 'EUC-JP') . "\n"; 27print mb_stripos($euc_jp, "\xC6\xFC\xCB\xDC\xB8\xEC", 15, 'EUC-JP') . "\n"; 28print mb_stripos($euc_jp, '0', 15, 'EUC-JP') . "\n"; 29print mb_stripos($euc_jp, 3, 15, 'EUC-JP') . "\n"; 30print mb_stripos($euc_jp, 0, 15, 'EUC-JP') . "\n"; 31 32 33// Negative offset 34echo "== NEGATIVE OFFSET ==\n"; 35 36print mb_stripos($euc_jp, "\xC6\xFC\xCB\xDC\xB8\xEC", -15, 'EUC-JP') . "\n"; 37print mb_stripos($euc_jp, '0', -15, 'EUC-JP') . "\n"; 38print mb_stripos($euc_jp, 3, -15, 'EUC-JP') . "\n"; 39print mb_stripos($euc_jp, 0, -15, 'EUC-JP') . "\n"; 40print mb_stripos($euc_jp, 0, -43, 'EUC-JP') . "\n"; 41 42 43// Out of range - should return false 44print "== OUT OF RANGE ==\n"; 45 46$r = mb_stripos($euc_jp, "\xC6\xFC\xCB\xDC\xB8\xEC", 40, 'EUC-JP'); 47($r === FALSE) ? print "OK_OUT_RANGE\n" : print "NG_OUT_RANGE\n"; 48$r = mb_stripos($euc_jp, '0', 40, 'EUC-JP'); 49($r === FALSE) ? print "OK_OUT_RANGE\n" : print "NG_OUT_RANGE\n"; 50$r = mb_stripos($euc_jp, 3, 40, 'EUC-JP'); 51($r === FALSE) ? print "OK_OUT_RANGE\n" : print "NG_OUT_RANGE\n"; 52$r = mb_stripos($euc_jp, 0, 40, 'EUC-JP'); 53($r === FALSE) ? print "OK_OUT_RANGE\n" : print "NG_OUT_RANGE\n"; 54$r = mb_stripos($euc_jp, "\xC6\xFC\xCB\xDC\xB8\xEC", -3, 'EUC-JP'); 55($r === FALSE) ? print "OK_OUT_RANGE\n" : print "NG_OUT_RANGE\n"; 56$r = mb_stripos($euc_jp, '0', -3, 'EUC-JP'); 57($r === FALSE) ? print "OK_OUT_RANGE\n" : print "NG_OUT_RANGE\n"; 58$r = mb_stripos($euc_jp, 3, -3, 'EUC-JP'); 59($r === FALSE) ? print "OK_OUT_RANGE\n" : print "NG_OUT_RANGE\n"; 60$r = mb_stripos($euc_jp, 0, -3, 'EUC-JP'); 61($r === FALSE) ? print "OK_OUT_RANGE\n" : print "NG_OUT_RANGE\n"; 62 63 64// Non-existent 65echo "== NON-EXISTENT ==\n"; 66 67$r = mb_stripos($euc_jp, "\xB4\xDA\xB9\xF1\xB8\xEC", 0, 'EUC-JP'); 68($r === FALSE) ? print "OK_STR\n" : print "NG_STR\n"; 69$r = mb_stripos($euc_jp, "\n", 0, 'EUC-JP'); 70($r === FALSE) ? print "OK_NEWLINE\n" : print "NG_NEWLINE\n"; 71 72 73// EUC-JP - No encoding parameter 74echo "== NO ENCODING PARAMETER ==\n"; 75 76mb_internal_encoding('EUC-JP') or print("mb_internal_encoding() failed\n"); 77 78print mb_stripos($euc_jp, "\xC6\xFC\xCB\xDC\xB8\xEC", 0) . "\n"; 79print mb_stripos($euc_jp, '0', 0) . "\n"; 80print mb_stripos($euc_jp, 3, 0) . "\n"; 81print mb_stripos($euc_jp, 0, 0) . "\n"; 82 83$r = mb_stripos($euc_jp, "\xB4\xDA\xB9\xF1\xB8\xEC", 0); 84($r === FALSE) ? print "OK_STR\n" : print "NG_STR\n"; 85$r = mb_stripos($euc_jp, "\n", 0); 86($r === FALSE) ? print "OK_NEWLINE\n" : print "NG_NEWLINE\n"; 87 88// EUC-JP - No offset and encoding parameter 89echo "== NO OFFSET AND ENCODING PARAMETER ==\n"; 90 91mb_internal_encoding('EUC-JP') or print("mb_internal_encoding() failed\n"); 92 93print mb_stripos($euc_jp, "\xC6\xFC\xCB\xDC\xB8\xEC") . "\n"; 94print mb_stripos($euc_jp, '0') . "\n"; 95print mb_stripos($euc_jp, 3) . "\n"; 96print mb_stripos($euc_jp, 0) . "\n"; 97 98$r = mb_stripos($euc_jp, "\xB4\xDA\xB9\xF1\xB8\xEC"); 99($r === FALSE) ? print "OK_STR\n" : print "NG_STR\n"; 100$r = mb_stripos($euc_jp, "\n"); 101($r === FALSE) ? print "OK_NEWLINE\n" : print "NG_NEWLINE\n"; 102 103echo "== INVALID STRINGS ==\n"; 104 105// Previously, mb_stripos would internally convert invalid byte sequences to the 106// mb_substitute_char BEFORE performing search 107// So invalid byte sequences would match the substitute char, both from haystack 108// to needle and needle to haystack 109 110mb_substitute_character(0x25); // '%' 111var_dump(mb_stripos("abc%%", "\xFF", 0, "UTF-8")); // should be false 112var_dump(mb_stripos("abc\xFF", "%", 0, "UTF-8")); // should be false 113 114// However, invalid byte sequences can still match invalid byte sequences: 115var_dump(mb_stripos("abc\x80\x80", "\xFF", 0, "UTF-8")); 116var_dump(mb_stripos("abc\xFF", "c\x80", 0, "UTF-8")); 117 118 119?> 120--EXPECT-- 121String len: 43 122== POSITIVE OFFSET == 12310 1240 1253 1260 12734 12830 12933 13030 131== NEGATIVE OFFSET == 13234 13330 13433 13530 1360 137== OUT OF RANGE == 138OK_OUT_RANGE 139OK_OUT_RANGE 140OK_OUT_RANGE 141OK_OUT_RANGE 142OK_OUT_RANGE 143OK_OUT_RANGE 144OK_OUT_RANGE 145OK_OUT_RANGE 146== NON-EXISTENT == 147OK_STR 148OK_NEWLINE 149== NO ENCODING PARAMETER == 15010 1510 1523 1530 154OK_STR 155OK_NEWLINE 156== NO OFFSET AND ENCODING PARAMETER == 15710 1580 1593 1600 161OK_STR 162OK_NEWLINE 163== INVALID STRINGS == 164bool(false) 165bool(false) 166int(3) 167int(2) 168
